Templates A, B and D

Three fill-in video prompts, written for this guide. Fill every slot, delete unneeded blocks, then validate. Templates A and B are for shots where nobody speaks a principal line; speaking shots use Templates C, E and F (see Speaking shots and lip sync → Templates C, E and F). Worked prompts: see Video with Seedance → How to read a recipe.

Template A: one continuous shot, references only

Settings: omni-reference; duration = screen time × 1.5–2, plus 0.5–1 s handles; the master's aspect; a 480p draft, then its 1080p finish; audio on; images: sheet, close-up, plate, product, prop.

SCENE CONTEXT: {the action, the place, the time and each person's position, in one or two sentences}.
ACTIVE REFERENCES: @image1 is {ALIAS}: {age} {build or role}, {their state now}, {distinctive features}, {what is in each hand}. This sheet gives identity and wardrobe and nothing else: the face and hair come from the close-up panel alone, the wardrobe and body from the full-body panels alone; the front panel's cropped head is never repeated; leave out its grey ground, panel edges, layout and flat light. Match it exactly. @image2 is {ALIAS}'S FACE: the identity close-up, used for the face alone; leave out its grey ground and flat light. @image3 is {PLACE}: its architecture, materials, landmarks and light, nothing else; the set holds exactly as pictured for the whole shot; it does not decide where the camera stands. @image4 is {PRODUCT}: its exact shape, colour, cap and label; where these words and the picture differ, follow the picture; it is not a start frame.
LOCATION MAP: From the lens: foreground {…} at {m}; midground {…} at {m}; background {…} from {m}, with {named details and their counts}. Left and right as the camera sees them; the camera keeps to the {side} of the line. The light comes from {source and side}. Live motion: {as the plan counts it; each with its cause, direction and rate}.
FIRST FRAME AND BLOCKING (in motion from frame one): {size}. {ALIAS} is {position in metres, facing}, already {mid-action}; left hand {…}; right hand {…}. EYELINE — {ALIAS}: looks {frame-left | frame-right} of the lens, {just past it | well wide of it}, at {target}, {level | a little up | a little down}, and stays on it through the move.
FORMAT MODE: A single unbroken shot with no cuts of its own. {Real time, with no slow motion and no speed ramps. | Slow motion: captured fast, shown at 24 fps.}
OPTICS: {size} on a {°} diagonal field of view throughout; {what only this lens shows}; the lens holds steady.
CAMERA: {rig} at {height} m, {distance} m from {subject}{; handheld: the operator's shoulder and breath, a fraction late on the turns, a small overshoot and a correction at each stop}. {the film's camera-law clause}. When {trigger} at about {t} s, one {pace} {move} {direction} of {dose} over {n} s, seen as {the framing change}, {travelling | staying in place and panning}{, passing {one soft foreground element}}. It opens on {start composition} and arrives on {landing composition}, {its job}. {The cut comes in motion, so the move carries through the last frame. | The cut comes after the beat, so the move eases to rest in the last second.}
ACTION / TIMELINE: From the first frame {stage 1 and its visible end state}; by about {a} s {stage 2}; at {duration} s {final state}. {Reaction, {overlapping | half a beat late | instant | held, then late}: {cause at t1}; at {t1 + latency} {one visible change}, which then holds.}
PERFORMANCE — {ALIAS}: {goal, obstacle and tactic as verbs aimed at the partner}. The eyes stay on {target}; the pupils move; one blink at about {t} s, where the thought turns. {one involuntary detail}. {Skin: {2–3 conditions of the day}.} {emotion}, never {overshoot}.
PHYSICS: {each material as cause and effect, with its weight}; each move gathers, travels and comes to rest: {the settle}. {the cadence: a 180-degree shutter, so fast movement blurs and the holds stay sharp}.
TEXTURE: {2–4 materials that show wear: one sign per garment, use marks where hands go}; the frame sits off-centre; the background falls into darkness past {n} m.
LIGHTING: One {quality} key from {position, seen or placed by the scene's first shot}, {Kelvin} K; {fill, or the unseen walls kept dark}; exposed for {priority}. {As the camera arcs, the key moves from {…} to {…}.} {how each material takes the key}. No surface lights itself.
AUDIO: Sound effects only. {the space and the listener's distance}; {the ambience bed}, with steady room tone from first frame to last; {t} s {sound, surface, source}; {t} s {sound}; off screen: {one layer, its side and distance}. No music and no voices; nobody speaks and all mouths stay shut. Nothing rises, swooshes, stings or fades at the start, at a cut or at the end. No subtitles and no on-screen text.
STYLE: {the film's look line, word for word, with the scene's capture as visible results}.
POSITIVE LOCKS: Exactly {counts}. {the only readable text}. The set stays exactly as in {PLACE}. {Only where the plan names the hero-colour option: {colour} is the richest colour in every frame.} The clip is wrong if {failure sentence}.

A named lock instead of a move, only where the plan names the exception:

CAMERA: Locked off on a tripod at {height} m, {distance} m from {subject}. The camera stays still the whole time: no zoom and no push-in. {what does not move}. Life in the frame: {one small, visible change every 1–2 s}.

With an internal frame: add the start image (and end image), drop every @ number, name each reference by its content, and add after ACTIVE REFERENCES:

FRAMES: The clip begins exactly on the opening frame{ and ends on the closing frame}. From the first instant {ALIAS} matches the opening frame exactly: {face, hair, wardrobe, hands}.

Template B: several shots in one generation

Settings: as Template A; the duration is the sum of the slots; neighbours differ in size, angle and action.

SCENE CONTEXT: {…}.
ACTIVE REFERENCES: {one role line per reference; each tag declared once, then its alias}.
LOCATION MAP: {layers from the lens, each at its distance}. FIXED GEOGRAPHY, every shot: {sides, doors and directions relative to the camera}. Live motion: {as the plan counts it}.
FIRST FRAME AND BLOCKING (in motion from frame one): {shot 1's first frame: who, where, doing what}. EYELINE — {ALIAS 1}: {side of the lens, target, height}; {ALIAS 2} looks to the other side.
FORMAT MODE: Exactly {n} shots and {n−1} cuts, timed below. The clip cuts at these points and nowhere else. Real time.
OPTICS: Each shot sets its own field of view below; inside a shot the lens holds steady.
Shot 1 (0.0–{a} s) — {SIZE}, {°} diagonal: CAMERA {one dosed move with its trigger and framing change, or the named lock}. {action}. BACKGROUND: {what stands behind each visible person; each extra's task, tempo and direction}. Hands: {inventory}. {CUT TYPE} CUT.
Shot 2 ({a}–{b} s) — {SIZE}, {°} diagonal: CAMERA {…}. {who is where now}. {action}. BACKGROUND: {…}. {CUT TYPE} CUT.
Shot {n} ({…}–{duration} s) — {SIZE}, {°} diagonal: CAMERA {…}. {action}. {how the shot ends, so the planned cut works}.
PERFORMANCE — {ALIAS 1}: {task, eye life, involuntary detail, reaction type and latency, clamp}. {ALIAS 2}: {in silent shots: silent in this beat and listening, eyes on {ALIAS 1}'s eyes; one change when the key word lands; otherwise still}.
PHYSICS: {…; each move gathers, travels and comes to rest}.
TEXTURE: {2–4 worn materials that recur across the shots}; frames off-centre.
LIGHTING: {one setup, written once and held in every shot; the key seen or placed in shot 1}.
AUDIO: {as Template A, with the sounds timed across the shots}.
STYLE: {the film's look line}.
POSITIVE LOCKS: {n} shots and {n−1} cuts, exactly. Exactly {people}. {wardrobe, props and time of day} in every shot. KEEP CONSISTENT: 1. {…} 2. {…} {8–14 short rules, one per thing this clip is likely to get wrong}.

The count-locked variant: FORMAT MODE and each header change as below; POSITIVE LOCKS repeats the count.

FORMAT MODE: {n} shots and {n−1} cuts, no more and no fewer, with no added inserts. The clip cuts at these points and nowhere else. Real time.
Shot 1 — {SIZE}, {°} diagonal, very short, about a second:

Template D: the extension paragraph

Settings: a forward extension, the original anchors still attached; straight after SCENE CONTEXT.

CONTINUITY: This clip carries straight on from the last one. Its first frame is the last clip's final frame: {what that frame shows}. The motion goes on without a pause: the same people, wardrobe, place, light and camera operator, and any held pose starts moving by 0.25 s. The light runs on across the join. Anything new enters only after the join.
TEXTURE: {the same worn materials as the last clip}.

Sources

• Higgsfield's prompt builder (the block order and anchor line); @adilinthewildtempo (BACKGROUND, KEEP CONSISTENT); JOEY, @joeysnoisygroup (the continuity paragraph); Higgsfield Academy (the count-locked form); Higgsfield's ANERNEQ (texture and the anti-render frame).